ChatGPT-5 users on social media platforms were critical of the update, and OpenAI CEO Sam Altman responded to the complaints, pledging improvements.
ChatGPT users flooded social media forums with negative comments about the platform’s latest model released on Thursday, saying the upgrade was overhyped by OpenAI CEO and co-founder Sam Altman.
Users on Reddit criticized the company’s ChatGPT-5 model, saying it had a more restrictive rate limit and lacked the ability to switch to previous AI models, with some threatening to switch to competing AI platforms. Altman responded to the feedback in an X post:
“We are going to double GPT-5 rate limits for ChatGPT Plus users as we finish rollout. We will let Plus users choose to continue to use 4o. We will watch usage as we think about how long to offer legacy models for.”
Altman called GPT-5 “a major upgrade over GPT-4” and a “significant step” along the road to artificial general intelligence (AGI) during Thursday’s official rollout.
Negative user feedback of the latest ChatGPT release comes amid growing competition from open-weighted, open-source and decentralized AI platforms that are less capital-intensive and are taking market share of large, centralized players in consumer AI applications.
Related: US government announces ChatGPT integration across agencies
Nimble AI project eats away at consumer AI market share
DeepSeek, an artificial intelligence company, released a new open-weighted large-language model in April called Prover V2. The LLM is tailored for use in mathematics.
Open-weight AI models allow anyone to download and use the software, but leave critical aspects like training data out of the publicly available information.
The added transparency is a step below full open-source code, in which users have full access to the codebase and can fully produce the software from that code.
However, open-weight software is more transparent than the centralized model in which the code is a black box and users are left in the dark.
The company also released an upgrade of DeepSeek R1, a more general-purpose open-weighted AI model, in May. DeepSeek R1-0528 functioned on par with leading AI models, including OpenAI-o3, according to benchmarks from the company.
The DeepSeek app has been downloaded more than 75 million times since its release in January, according to Business of Apps.
Magazine: Cypherpunk AI: Guide to uncensored, unbiased, anonymous AI in 2025
